The study reveals that each individual plant possesses a statistically unique leaf appearance that can be discriminated using convolutional neural network (CNN) based deep learning, enabling a "plant face" recognition concept. Applications demonstrated include distinguishing leaves from the same species/cultivar, analyzing leaflet positional patterns on compound leaves, assessing bilateral symmetry, and detecting morphological differences linked to stem chirality, highlighting the encoding of genetic, environmental, and developmental information in leaf morphology.

The study demonstrates that the plastid chaperone CLPC2, but not its paralogue CLPC1, is essential for Arabidopsis responsiveness to microbial volatile compounds and for normal seed and seedling development. Loss of CLPC2 alters the chloroplast proteome, affecting proteins linked to growth, photosynthesis, and embryogenesis, while overexpression of CLPC2 mimics CLPC1 deficiency, highlighting distinct functional roles within the CLP protease complex.

The authors compiled and standardized published data on Rubisco dark inhibition for 157 flowering plant species, categorizing them into four inhibition levels and analyzing phylogenetic trends. Their meta‑analysis reveals a complex, uneven distribution of inhibition across taxa, suggesting underlying chloroplast microenvironment drivers and providing a new resource for future photosynthesis improvement efforts.

The study examined how polyploidy, hybridization, and incomplete lineage sorting affect phylogenetic reconstructions in the genus Packera, evaluating several published paralog‑processing pipelines. Results showed that the choice of orthology and paralog handling methods markedly altered tree topology, time‑calibrated phylogenies, biogeographic histories, and detection of ancient reticulation, underscoring the need for careful methodological selection alongside comprehensive taxon sampling.

The study evaluated whether integrating genomic, transcriptomic, and drone-derived phenomic data improves prediction of 129 maize traits across nine environments, using both linear (rrBLUP) and nonlinear (SVR) models. Multi-omics models consistently outperformed single-omics models, with transcriptomic data especially enhancing cross‑environment predictions and capturing genotype‑by‑environment interactions. The results highlight the added value of combining transcriptomics and phenomics with genotypes for more accurate and generalizable trait prediction in maize.